Warfarin is a mainstay of cardiovascular pharmacotherapy and one of the most frequently prescribed medications in the U.S. However, major bleeding is a frequent and potentially devastating complication of its use, affecting approximately 3% of patients annually. Serious warfarin-related bleeding most commonly occurs in the gastrointestinal (GI) tract, accounting for about 1/3 of major bleeds. This suggests proton-pump inhibitors (PPIs) might improve warfarin safety. This hypothesis is supported by our understanding of the pathophysiology of warfarin-related bleeding as well as by studies of other pro-hemorrhagic drugs. A benefit for PPI cotherapy would be greatest for the many warfarin users with other risk factors for GI bleeding. However, the hypothesis that PPIs prevent serious warfarin-related GI bleeding, although plausible, has not been tested. The available evidence is largely indirect and cannot, by itself, support recommending PPI cotherapy to warfarin patients. Indeed, contemporary anticoagulant guidelines do not mention PPIs. Histamine-2 receptor antagonists (H2RAs) also might be considered to prevent serious warfarin-related GI bleeding, although they may be less effective than PPIs. Novel oral anticoagulants, more convenient to use than warfarin, will account for a growing proportion of oral anticoagulant use. However, these also cause serious GI bleeding; some have risk greater than that of warfarin. Thus, PPIs may have the potential to markedly improve the safety of novel anticoagulants. We will conduct a large historical cohort study in Tennessee Medicaid and the national 5% Medicare sample assessing whether concurrent PPI/H2RA in oral anticoagulant users reduces risk of serious GI bleeding, critical data to improve the safety of these widely used medications. We will test two hypotheses: Aim 1: In Tennessee Medicaid warfarin users, concurrent PPI or H2RA use decreases risk of hospitalizations for gastroduodenal bleeding. Aim 2: In novel oral anticoagulant users in the Medicare sample, concurrent PPI or H2RA use decreases risk of hospitalizations for gastroduodenal bleeding. For each of these aims, we also will test whether the protective effect is greater for PPIs than for H2RAs as well as whether the absolute benefit varies according to number of GI bleeding risk factors.